Abstract A021: High-dimensional biopsy-derived proteomics identify tumor-intrinsic inflammatory phenotypes associated with short-term mortality in pancreatic ductal adenocarcinoma
Gerik W. Tushoski-Alemán, Alexandra J. Crespin, Sophia A. Rosales, Enrique A. Maduro, Grace R. Thompson, Maksymilian J. Pilecki, Patrick W. Underwood, Charles E. Middleton, Andrew A. Brock, Karthik Chandrasekaran, Dejana Braithwaite, Song Han, Tiffany Chua, Steven J. Hughes, Aleksey A. NovikovAbstract
Introduction:
Diagnostic biopsies represent the first opportunity to collect tumor tissue in pancreatic ductal adenocarcinoma (PDAC) and present an opportunity for early prognostication. We hypothesized that tumor-intrinsic inflammatory protein levels reflect disease biology and are associated with survival. Thus, we aimed to identify tumor-intrinsic inflammatory phenotypes from biopsies and test their association with 6-month mortality.
Methods:
Endoscopic ultrasound-guided fine-needle biopsies (EUS-FNB) were prospectively collected from 28 patients with solid pancreatic masses from June 2024 to October 2025, of which 24 patients were diagnosed with PDAC. Samples were profiled using an 80-plex inflammatory protein panel comprised of 35 cytokines, 26 chemokines, 6 growth factors, and 13 inflammatory mediators. Analytes were normalized to total protein, log-transformed, and used for unsupervised hierarchical clustering. The prespecified primary endpoint was 6-month mortality. Secondary analyses included overall survival, phenotype-defining proteins, compartment-level localization, and comparison of canonical driver mutation alteration frequency using next-generation sequencing.
Results:
The PDAC cohort consisted of 24 patients, 13 (54.2%) were male, 18 (75%) were > 65 years old, 9 (37.5%) with locally advanced, 10 (41.7%) with borderline resectable, and 5 (20.8%) with resectable disease. The cohort separated into two inflammatory phenotypes: Phenotype 1 (P1) (n=10) and Phenotype 2 (P2) (n=14) using unsupervised hierarchical clustering. Six-month mortality was 0/10 (0%) in P1 tumors versus 4/14 (28.6%) in P2 tumors (p = 0.11). P2 was associated with inferior overall survival (HR 7.14, 95% CI 0.89-57.27; log-rank p=0.0308). After adjusting for false discovery, HGF and FGF2 were enriched in P1 tumors, and Galectin-3 was enriched in P2 tumors. Next-generation sequencing of samples with sufficient material (P1, n = 8) (P2, n = 11) showed similar mutation alteration frequency among canonical PDAC drivers between P1 and P2: KRAS (100% vs 100%), TP53 (87.5% vs 81.8%), SMAD4 (25.0% vs 36.4%), and CDKN2A (12.5% vs 18.2%). Orthogonal validation showed compartment-level localization of phenotype-defining proteins in tissue, where P1 proteins FGF2 localized to fibroblasts and HGF localized to cancer cells, while the P2 protein (Galectin-3) localized to cancer cells.
Conclusions:
Our preliminary findings identified a tumor-intrinsic inflammatory phenotype from diagnostic EUS-FNB tissue that was enriched for higher 6-month mortality and inferior overall survival. Future research will focus on biologic and biomarker validation of these phenotypes in larger independent cohorts to assess their biologic and clinical relevance and support treatment decision-making.
Citation Format:
Gerik W. Tushoski-Alemán, Alexandra J. Crespin, Sophia A. Rosales, Enrique A. Maduro, Grace R. Thompson, Maksymilian J. Pilecki, Patrick W. Underwood, Charles E. Middleton, Andrew A. Brock, Karthik Chandrasekaran, Dejana Braithwaite, Song Han, Tiffany Chua, Steven J. Hughes, Aleksey A. Novikov. High-dimensional biopsy-derived proteomics identify tumor-intrinsic inflammatory phenotypes associated with short-term mortality in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Conference on Pancreatic Cancer: New Frontiers in Biology and Therapeutic Development; 2026 Sep 25-28; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(18_Suppl_2):Abstract nr A021.